MGT4202 : System Modeling & Decision

Making

Version 1.1 || (August 19, 2021)

Prepared by:

A B M Shahadat Hossain, PhD

Associate Professor of Applied Mathematics
University of Dhaka

Email:[email protected]
Preface
This lecturenotes aim to give you a summary of the course material. The
proofs and solutions shown in the lectures are not written down in the notes .
And most of the comments, whether or not helpful, from the lecture you can
not find neither. We will follow a text book throughout the course written
by Professor Hamdy A. Taha [1]. You should also see the optional books
mentioned in the course outline.[2, 3]
MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
Lec 01
Introduction

Preamble

This course is a basic introduction to important models and solu-

tion methods in Operations Research (OR).

Operations Research is concerned with the modeling and analysis

of complex decision problems that arise in the management of an
organization.

The basic theme in OR is optimization; that is, the goal is to

improve the efficiency of operations through detailed modeling and
analysis.

What is Operations Research?

Operations Research originated in Great Britain during World War

II to bring mathematical or quantitative approaches to bear on
military operations. Thus, Operations Research is concerned with
research on operations; and the word ”research” means that math-
ematical methods are used to conduct rigorous scientific analyses
of complex problems.

Institute: MGT, BUP Page# 2

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Since its birth in the 1940’s, OR has been widely recognized as an

important approach to decision-making in the management of all
aspects of an organization. (Consequently, OR is also referred to
as Management Science.) Hence, the word ”operations” should be
interpreted broadly as any operation that requires the efficient allo-
cation of limited resources. Specific areas of application include, for
example, manufacturing, transportation and distribution, finance,
marketing, health care, and telecommunications.

The mathematical methods most widely used in OR include math-

ematical programming, probability and statistics, and computer
science. Some areas of application of OR, such as inventory and
production control, queueing theory, scheduling theory, and simu-
lation have, in fact, grown into subdisciplines in their own right.

For an overview of the history and the impact of Operations Re-

search, read Chapter 1 of the text (by Hillier). In particular, Table
1.1 provides a good summary of applications of OR in industry.

Definitions of OR

“Operations research is the systematic application of quantitative

methods, techniques and tools to the analysis of problems involving
the operation of systems.” – Daellenbach and George, 1978

Institute: MGT, BUP Page# 3

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

“The application of the scientific method to the study of opera-

tions of large complex organizations or activities. It provides top
level administrators with a quantitative basis for decisions that will
increase the effectiveness of such organizations in carrying out their
basic purposes.” – Committee on OR National Research Council,
USA.

“Operations research is a scientific approach to problem-solving for

executive management.” – H M Wagner

These definitions imply a view of OR – being the collection of models and

methods that have been developed largely independent of one another.

As the discipline of operations research grew, numerous names

such as Operations Analysis, Systems Analysis, Decision Analysis,
Management Science, Quantitative Analysis, Decision Science were
given to it. This is because of the fact that the types of problems
encountered are always concerned with ‘effective decision’.

What is a Mathematical Model?

The majority of practical decision problems are described in very

vague terms. Therefore, a most-important step in a scientific or
quantitative analysis of a problem is to formulate a model that ad-
equately captures the essence of a problem. The result of such a
formulation, or an abstraction, is called a mathematical optimiza-
tion model.

Institute: MGT, BUP Page# 4

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Mathematical model uses mathematical equations and state-

ments to represent the relationships within the model.

Figure 1: Clssification of OR models

In a deterministic model all values used are known with certainty.

Institute: MGT, BUP Page# 5

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

In a Probabilistic model all values used are not known with cer-
tainty; and are often measured as probability values.

Garden problem: A home owner is in the process of starting a

backyard vegetable garden. The garden must take on a rectangular
shape to facilitate row irrigation. To keep critters out, the garden
must be fenced. The owner has enough material to build a fence
of length L = 100 ft. The goal is to fence the largest possible
rectangular area.
Solution: in lecture

Generally speaking, a mathematical optimization model has the

following typical components:

◦ a set of decision variables

◦ an objective function, expressed in terms of the decision vari-

ables, that is to be minimized or maximized

◦ a set of constraints that limit the possible values of the de-

cision

Two concrete examples are:

1. in production planning, one could be interested in setting a

production schedule (the decision variables) that minimizes
the total cost (the objective function) of operation over a
given planning horizon, for a given production capacity (the

Institute: MGT, BUP Page# 6

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

constraint); and

2. in investment management, one could be interested in

assembling a collection of securities (the decision variables)
that maximizes the expected yield (the objective function)
of a portfolio, under a limited budget (the constraint).

The OR Modeling Approach

In addition to the formulation of a model, the OR approach in-

volves several additional steps. The next step, of course, is to ap-
ply (or, sometimes, to develop) a solution method to derive a set
of values for the decision variables that ”optimizes” the objective
function subject to the given constraints. After having formulated
and solved a problem, the remaining steps then consist of model
validation and implementation.

Thus, the basic steps in the OR modeling approach are:

◦ Problem Identification

◦ Formulation

◦ Derivation of Optimal Solution(s)

◦ Model Validation

◦ Implementation

Institute: MGT, BUP Page# 7

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

It is important to realize that before an actual implementation,

one should cycle through the first four steps a sufficient number
of times so as to ensure that a reasonably valid model has been
achieved.

For a detailed discussion of these steps, read Chapter 2 of the text.

Figure 2: Methodology of OR

Institute: MGT, BUP Page# 8

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
Methods for Solving OR Models

In general, operations research models are solved using either

(i) analytical,

(ii) numerical, or

(iii) Monte Carlo method

Analytical (or deductive) method:

– In this method, classical optimization techniques such as cal-
culus, finite difference and graphs are used for solving an OR
model.

– The analytical methods are noniterative methods to obtain

an optimal solution of a problem.

Numerical (or iterative) method:

– When analytical methods fail to obtain the solution of a par-
ticular problem due to its complexity in terms of constraints
or number of variables, a numerical (or iterative) method is
used to find the solution.

– In this method, instead of solving the problem directly, a

general algorithm is applied for obtaining a specific numer-
ical solution. The numerical method starts with a solution
obtained by trial and error, and a set of rules for improving
it towards optimality.

Institute: MGT, BUP Page# 9

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Monte Carlo method:

– This method is based upon the idea of experimenting on
a mathematical model by inserting into the model specific
values of decision variables for a selected period of time un-
der different conditions and then observing the effect on the
criterion chosen.

– In this method, random samples of specified random variables

are drawn to know how the system is behaving for a selected
period of time under different conditions.

– The random samples form a probability distribution that rep-

resents the real-life system and from this probability distri-
bution, the value of the desired random variable can then be
estimated.

Applications of OR
Finance and Accounting
ˆ Dividend policies, investment and
ˆ Claim and complaint pro-
portfolio management, auditing,
cedure, and public ac-
balance sheet and cash cost flow
counting
analysis
ˆ Establishing costs for by-
ˆ Break-even analysis, capital bud-
products and developing
geting, allocation and control,
standard costs
and financial planning

Institute: MGT, BUP Page# 10

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
Marketing
ˆ Sales effort allocation and as-
ˆ Selection of product-mix, mar-
signment
keting and export planning
ˆ Launching a new product at
ˆ Advertising, media planning,
the best possible time
selection and effective packing
alternatives ˆ Predicting customer loyalty

Techniques and General Management

ˆ Making quality control more
ˆ Decision support systems and effective
MIS; forecasting ˆ Project management and
strategic planning

Govermment
ˆ Economic planning, natural ˆ Urban and housing problems
resources, social planning and ˆ Military, police, pollution con-
energy trol, etc.

OR Models in Practice
There is no unique set of problems that can be solved by using OR techniques.
Several OR techniques can be grouped into some basic categories as given
below. In this cours, a large number of OR techniques have been discussed in
detail.

Institute: MGT, BUP Page# 11

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Allocation models are used to allocate resources to activities so

as to optimize measure of effectiveness (objective function). Math-
ematical programming is the broad term for the OR techniques used
to solve allocation problems.

Inventory models deal with the problem of determination of how

much to order at a point in time and when to place an order. The
main objective is to minimize the sum of three conflicting inventory
costs:

Waiting line (or Queuing) models establish a trade-off between

costs of providing service and the waiting time of a customer in
the queuing system.

Competitive (Game Theory) models are used to characterize

the behaviour of two or more competitors (called players) compet-
ing to achieve their conflicting goals.

Network models are applied to the management (planning, con-

trolling and scheduling) of large-scale projects. PERT/CPM tech-
niques help in identifying delay and project critical path.

Dynamic programming models are used where a problem re-

quires optimization of multistage (sequence of interrelated deci-
sions) decision processes.

Institute: MGT, BUP Page# 12

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Markov-chain models are used for analyzing a system which

changes over a period of time among various possible outcomes
or states.

Simulation models are used to evaluate alternative courses of ac-

tion by experimenting with a mathematical model of the problems
with random variables.

Decision analysis models deal with the selection of an optimal

course of action given the possible payoffs and their associated

Course Coverage

OR models are generally classified as either deterministic or

stochastic. A deterministic model is one in which all problem
parameters are assumed to be known with certainty; whereas a
stochastic model is one in which some of the parameters are as-
sumed to be random.

The emphasis in this (short) course is on deterministic models. In

particular, we will focus on two topics, namely linear programming
and dynamic programming. Linear programming refers to the for-
mulation and solution of mathematical optimization models that
have a linear objective function and a set of linear constraints; and
dynamic programming refers to a versatile mathematical method
that can be used to solve optimization problems that involve a

Institute: MGT, BUP Page# 13

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

sequence of interrelated decisions, usually over time. These two

topics are chosen mainly for their wide range of applicability.

Our coverage corresponds to most of the chapters in the text.

Despite our limited scope, it is hoped that the course will help
you acquire a solid understanding of how complex problems can be
modeled and solved with these methods.

Quiz

True (T) or False (F)

1. Operations research approach does not adequately meet the needs

of business and industry.

2. Operations research is a scientific approach to problem solving for

executives.

3. Simulation models are comparatively less flexible than any mathe-

matical models.

4. All decisions are made either on the basis of intuition or by using

more formal approaches.

5. Operations research practioners can predict about the future

events.

Institute: MGT, BUP Page# 14

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
Lec 02
Linear Programming (LP)

Preamble

A linear program is a mathematical optimization model that has a

linear objective function and a set of linear constraints.

To provide a quick overview, we describe below what is known as

the product-mix problem.

The Product-Mix Problem

A company manufactures two models of a product, which we call

the regular model and the enhanced model. Two resources, A and
B, are needed for the manufacturing of units of this product; and
the detailed resource requirements are given in the following table:

Regular Model Enhanced Model

Resource A 3 4
Resource B 2 3

This means, for example, 3 units of Resource A and 2 units of

Resource B are consumed in the manufacturing of each unit of
the regular model. The resources could represent hours of labor,
amounts of raw material, available electrical power, and so on.

Institute: MGT, BUP Page# 15

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Suppose further that the company has 650 units of Resource A and
500 units of Resource B available per week, and that the net profits
for selling units of the regular model and the enhanced model are
given respectively as 5 and 7 per unit. The question of interest
is: What are the optimal weekly production levels for these two
models?

To formulate this problem, we first define a set of decision variables.

Let

xr = weekly production level for the regular model,

xe = weekly production level for the enhanced model,

Clearly, these variables should be nonnegative, as they represent

physical quantities.

Next, suppose our goal is to maximize weekly total profit. Then,

in terms of these decision variables, we have:

Weekly Total Profit = 5xr + 7xe ;

and we wish to choose a pair of values for xr and xe that yields the
highest possible weekly total profit.

Institute: MGT, BUP Page# 16

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

However, from the table above, we see that there are limitations or
constraints on the possible values of these variables. Specifically,
we have:

Weekly Requirement for Reasource A = 3xr + 4xe ,

Weekly Requirement for Reasource B = 2xr + 3xe ,

Therefore, the values of xr and xe should be such that these re-

quirements do not exceed 650 and 500, respectively.

Our problem can now be stated formally as:

Maximize 5xr + 7xe

Subject to: 3xr + 4xe ≤ 650

2xr + 3xe ≤ 500
xr ≥ 0, xe ≥ 0
What we have just formulated is called a linear program (LP).
In this example, it has

⇒ two decision variables, xr and xe ,

⇒ an objective function, 5xr + 7xe , and

⇒ a set of four constraints.

The objective function is to be maximized subject to the specified

constraints on the decision variables. It is customary to refer to
the first group of constraints,

3xr + 4xe ≤ 650

2xr + 3xe ≤ 500,

Institute: MGT, BUP Page# 17

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

as the functional constraints; and the remaining group,

xr ≥ 0, xe ≥ 0

as the nonnegativity constraints.

Notice that each term in the objective function has the form cx,
where c is a constant and x is a variable.
In other words, the objective function is linear in the decision vari-
ables xr and xe .
Notice further that the left-hand-side expressions in all four con-
straints are also linear. This is why we call the above problem a
linear program.
If a term such as 3x2 appears in a formulation, then the resulting
problem is said to be nonlinear.

Genearal Notation of a LP

More generally, a linear program with m functional constraints and

n decision variables (or ”activities”), x1 , x2 , . . . , xn , has the following
general format:

Institute: MGT, BUP Page# 18

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Maximize or Minimize c1 x1 + c2 x2 + · · · + cn xn

Subject to: a11 x1 + a12 x2 + . . . a1n xn ≤, =, ≥ b1

a21 x1 + a22 x2 + . . . a2n xn ≤, =, ≥ b2
.. .. .. ..
. . . .
am1 x1 + am2 x2 + . . . amn xn ≤, =, ≥ bm
x1 , x2 , . . . , xn ≥ 0

where the c j ’s, the ai j ’s, and the bi ’s are given parameters.
Notice that three types of constraints are allowed: ”≤”, ”=”,
or ”≥”. Strict inequalities, i.e., ”<” or ”>”, are not allowed in a
linear program, because they may lead to ill-defined problems (e.g.,
maximize x subject to x < 1).

Assumptions in a LP Formulation
In any formulation effort, one has to make simplifying assumptions. We
now describe more formally a number of important assumptions in a linear-
programming formulation:

Proportionality: The total contribution of any variable (or

activity), say x, to either the objective function or a constraint is
proportional to x; i.e., the total contribution assumes the form cx,
where c is a constant.

In the product-mix problem, for example, the total contributions to

the net profit from the sale of the regular model and the enhanced
model are assumed to be given, respectively, by 5xr and 7xe .

Institute: MGT, BUP Page# 19

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Additivity: The combined contribution from different variables,

say x1 and x2 , to either the objective function or a constraint
is the sum of their individual contributions; i.e., the combined
contribution assumes the form c1 x1 + c2 x2 .

In the product-mix problem, for example, the total net profit equals
the sum of 5xr and 7xe .

Divisibility: Every decision variable can assume fractional values;

i.e., all variables are continuous.

This assumption may not be valid in the product-mix problem,

since the product under consideraton could be, say, cars.

Certainty: All input parameters of the problem (i.e., the c j ’s, the
ai j ’s, and the bi ’s) are assumed to be known with certainty, i.e.,
deterministic.

For a more thorough discussion of these assumptions, read Section

3.3 of the text.

Other assumptions that are implicit in our formulation of the

product-mix problem include:

◦ All manufactured products can be sold.

◦ The setup cost for production is ignored.

◦ Decisions are made on a weekly basis.

Institute: MGT, BUP Page# 20

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

In general, in formulating a problem, one should strive for a good

balance between realism and mathematical tractability.

The product-mix problem can be solved by the Simplex method,

which is an efficient mathematical procedure for solving linear
programs. It can be shown that the optimal solution to our
product-mix problem is to produce a ”mix” of: 0 regular model
and 162.5 enhanced models. This solution achieves an objective-
function value of 1137.5.

This optimal solution was generated by LINDO (https://www.

lindo.com/), a software for solving mathematical programs. You
my download and install the software.

Institute: MGT, BUP Page# 21

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
Chapter 02
Modeling with Linear Programming (LP)

2.1 Two-Variable LP Model

Example 2.1-1 : The Reddy Mikks Company

Reddy Mikks produces both interior and exterior paints from two
raw materials, M1 and M2. The following table provides the basic
data of the problem:

The daily demand for interior paint cannot exceed that for exte-
rior paint by more than 1 ton. Also, the maximum daily demand
for interior paint is 2 tons. Reddy Mikks wants to determine the
optimum (best) product mix of interior and exterior paints that
maximizes the total daily profit.

Solution: in lecture.

Any values of decision variables (x1 , x2 , . . . , xn ) that satisfy all five

constraints constitute a feasible solution. Otherwise, the solu-
tion is infeasible.

The feasible solution that maximizes/minimizes the objective func-

tion is called the optimum soluion.

Institute: MGT, BUP Page# 22

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
2.2 Graphical LP Solution

The graphical solution includes two steps:

1. Determination of the feasible solution space.

2. Determination of the optimum solution from among all the

points in the solution space.

Example 2.1-1 : The Reddy Mikks Company

Solve the Reddy Mikks model Graphically.

Solution: in lecture.

Remark:
The optimum solution of an LP, when it exists, is always associated
with a corner point of the solution space, thus limiting the search
for the optimum from an infinite number of feasible points to a
finite number of corner points.

Institute: MGT, BUP Page# 23

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
2.2.2 Solution of a Minimization Problem

Example 2.2-2 : Diet Problem

Ozark Farms uses at least 800 lb of special feed daily. The special
feed is a mixture of corn and soybean meal with the following
compositions:

The dietary requirements of the special feed are at least 30%

protein and at most 5% fiber. The goal is to determine the daily
minimum-cost feed mix.

Solution: in lecture.

Remarks:

◦ at least corresponds to ≥

◦ at most corresponds to ≤

Institute: MGT, BUP Page# 24

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
2.4 LP Applications
ˆ This section presents realistic LP models in which the definition of the
variables and the construction of the objective function and the con-
straints are not as straightforward as in the case of the two-variable
model.

The areas covered by these applications include the following:

1. Investment.

2. Production planning and inventory control.

3. Workforce planning.

4. Urban development planning.

5. Oil refining and blending.

Remark:
Here we will discuss each model is detailed, and its optimum solu-
tion is interpreted.

2.4.1 Investment
ˆ Multitudes of investment opportunities are available to today’s investor.

ˆ Examples of investment problems are capital budgeting for projects,

bond investment strategy, stock portfolio selection, and establishment
of bank loan policy.

ˆ In many of these situations, LP can be used to select the optimal mix of

opportunities that will maximize return while meeting requirements set

Institute: MGT, BUP Page# 25

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
by the investor and the market.

Example 2.4-1 : Bank Loan Model

Bank One is in the process of devising a loan policy that involves a
maximum of $12 million. The following table provides the pertinent
data about available loans.

Type of loan Interest rate Bad-debt ratio

Personal .140 .10
Car .130 .07
Home .120 .03
Farm .125 .05
Commercial .100 .02

Bad debts are unrecoverable and produce no interest revenue.

Competition with other financial institutions dictates the allocation
of at least 40% of the funds to farm and commercial loans.
To assist the housing industry in the region, home loans must
equal at least 50% of the personal, car, and home loans. The
bank limits the overall ratio of bad debts on all loans to at most
4%.

Solution: in lecture.

Remarks: A subtle assumption in the preceding formulation is that

all loans are issued at approximately the same time. This allows
us to ignore differences in the time value of the funds allocated to

Institute: MGT, BUP Page# 26

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

the different loans.

2.4.2 Production Planning and Inventory Control

ˆ There is a wealth of LP applications in the area of production planning
and inventory control.

ˆ This section presents three examples. The first deals with production
scheduling to meet a single-period demand.

ˆ The second deals with the use of inventory in a multiperiod production

system to meet future demand, and

ˆ the third deals with the use of inventory and worker hiring/firing to
“smooth” production over a multiperiod planning horizon.

Example 2.4-2 : Single-Period Production Model

In preparation for the winter season, a clothing company is manu-
facturing parka and goose overcoats, insulated pants, and gloves.
All products are manufactured in four different departments: cut-
ting, insulating, sewing, and packaging.
The company has received firm orders for its products.
Th e contract stipulates a penalty for undelivered items.
Devise an optimal production plan for the company based on the
following data:

Institute: MGT, BUP Page# 27

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Time per unit(hr)

Department Parka Goose Pants Gloves Capacity(hr)

Cutting .30 .30 .25 .15 1000

Insulating .25 .35 .30 .10 1000
Sewing .45 .50 .40 .22 1000
Packaging .15 .15 .1 .05 1000
Demand 800 750 600 500
Unit profit $30 $40 $20 $10
Unit penalty $15 $20 $10 $8

Solution: in lecture.

Remarks: A subtle assumption in the preceding formulation is that

all loans are issued at approximately the same time. This allows
us to ignore differences in the time value of the funds allocated to
the different loans.

2.4.3 Workforce Planning

ˆ The idea of redefining the start of a shift to accommodate fluctuation
in demand can be extended to other operating environments as well.

ˆ The following example deals with the determination of the minimum

number of buses needed to meet rush-hour and offhour transportation
needs.

Example 2.4-5 : (Bus Scheduling Model)

Progress City is studying the feasibility of introducing a mass-
transit bus system to reduce in-city driving. The study seeks

Institute: MGT, BUP Page# 28

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

the minimum number of buses that can handle the transporta-

tion needs.
After gathering necessary information, the city engineer noticed
that the minimum number of buses needed fluctuated with time of
the day, and that the required number of buses could be approxi-
mated by constant values over successive 4-hr intervals.
The following figure (Figure 2.11) summarizes the engineer’s find-
ings. To carry out the required daily maintenance, each bus can
operate only 8 successive hours a day.

Solution: in lecture.

Remarks: The optimal solution obtained by Excel solver calls for

scheduling 26 buses (compared with 30 buses when the three tra-

Institute: MGT, BUP Page# 29

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

ditional shifts are used).

2.4.4 Urban Development Planning

ˆ Urban planning deals with three general areas:

(1) building new housing developments,

(2) upgrading inner-city deteriorating housing and recreational areas,
and
(3) planning public facilities (such as schools and airports).

ˆ The constraints associated with these projects are both economic (land,
construction, and financing) and social (schools, parks, and income
level).

ˆ The objectives in urban planning vary. In new housing developments,

profit is usually the motive for undertaking the project.

ˆ In the remaining two categories, the goals involve social, political, eco-
nomic, and cultural considerations.

ˆ Indeed, in a publicized case in 2004, the mayor of a city in Ohio wanted

to condemn an old area of the city to make way for a luxury housing
development. The motive was to increase tax collection to help alleviate
budget shortages.

ˆ The example presented in this section is fashioned after the Ohio case.

Example 2.4-6 : (Urban Renewal Model)

The city of Erstville is faced with a severe budget shortage. Seeking
a long-term solution, the city council votes to improve the tax base
by condemning an inner-city housing area and replacing it with a

Institute: MGT, BUP Page# 30

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

modern development.
The project involves two phases:
(i) demolishing substandard houses to provide land for the new
development and

(ii) building the new development.

The following is a summary of the situation.
1. As many as 300 substandard houses can be demolished. Each
house occupies a .25-acre lot. The cost of demolishing a
condemned house is $2000.

2. Lot sizes for new single-, double-, triple-, and quadruple-

family homes (units) are .18, .28, .4, and .5 acre, respec-
tively. Streets, open space, and utility easements account
for 15available acreage.

3. In the new development, the triple and quadruple units ac-

count for at least 25total. Single units must be at least 20

4. The tax levied per unit for single, double, triple, and quadru-
ple units is $1000, $1900, $2700, and $3400, respectively.

5. The construction cost per unit for single-, double-, triple-,

and quadruple-family homes is $50,000, $70,000, $130,000,
and $160,000, respectively.

6. Financing through a local bank is limited to $15 million.

How many units of each type should be constructed to
maximize tax collection?

Solution: in lecture.

Institute: MGT, BUP Page# 31

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Remarks: Linear programming does not automatically guarantee

an integer solution, and this is the reason for rounding the contin-
uous values to the closest integer.

Institute: MGT, BUP Page# 32

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
Chapter 03
The Simplex Method and Sensitivity Analysis

3.1 LP Model in Equation Form

The development of the simplex method computations is facilitated

by imposing two requirements on the LP model:

1. All the constraints are equations with nonnegative right-hand

side.

2. All the variables are nonnegative.

Converting inequalities into equations with nonnegative

right-hand side.

1) To convert a≤ -inequality to an equation, a nonnegative slack vari-

able is added to the left-hand side of the constraint. For example,
the M1-constraint of the Reddy Mikks model (Example 2.1-1)
is converted into an equation as

6x1 + 4x2 + s1 = 24, s1 ≥ 0

The nonnegative variable s1 is the slack (or unused amount) of

resource M1.

Institute: MGT, BUP Page# 33

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

2) Conversion from (≥) to (=) is achieved by subtracting a nonneg-

ative surplus variable from the left-hand side of the inequality. For
example, in the diet model (Example 2.2-2), the surplus variable
S1 (≥) converts the (≥) feed mix constraint to the equation

x1 + x2 − S1 = 800, S1 ≥ 0

The amount of S1 represents the excess tons of the mix over the
required minimum (= 800 tons).

The only remaining requirement is for the right-hand side of the

resulting equation to be nonnegative. The requirement can be
satisfied simply by multiplying both sides of the equation by −1, if
necessary.

Dealing with unrestricted variables

The use of an unrestricted variable in an LP model is demonstrated

in the multiperiod production smoothing model of Example 2.4-4,
where the unrestricted variable Si represents the number of workers
hired or fired in period i.

In the same example, the unrestricted variable is replaced by two

nonnegative variables by using the substitution

Si = Si− − Si+ , Si− ≥ 0, Si+ ≥ 0

In this case, Si− represents the number of workers hired and Si+ the
number of workers fired.

Institute: MGT, BUP Page# 34

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
Transition From Graphical to Algebraic Solution

Example 3.2-1 Consider the following LP with two variables:

Maximize z = 2x1 + 3x3

s/t
2x1 + x2 ≤ 4

x1 + 2x2 ≤ 5

x1 , x2 ≥ 0

Solve this LP by Graphical Method as well as by Algebraic Method.

Solution: in lecture.

Transition from graphical to algebraic solution can be represented

by the following (FIGURE 3.1, pp.101):

Institute: MGT, BUP Page# 35

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

In the algebraic solution space (defined by m×n equations, m < n),

basic solutions correspond to the corner points in the graphical
solution space.
⇒ They are determined by setting n − m variables equal to zero and
solving the m equations for the remaining m variables, provided the
resulting solution is unique.
⇒ This means that the maximum number of corner points is

n n!
Cm =
m!(n − m)!

As with corner points, the basic feasible solutions completely define

the candidates for the optimum solution in the algebraic solutions
space.

3.3 The Simplex Method

Rather than enumerating all the basic solutions (corner points) of

the LP problem (as we did in Section 3.2), the simplex method
investigates only a “select few” of these solutions.

Example 3.3-1 :
Recall the Reddy Mikks model:

Institute: MGT, BUP Page# 36

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Maximize z = 5x1 + 4x2

s/t
6x1 + 4x2 ≤ 24 (Raw material M1)

x1 + 2x2 ≤ 6 Raw material M2

−x1 + x2 ≤ 1 Market Limit

x2 ≤ 2 Demand limit

x1 , x2 ≥ 0

Solve this model using the Simplex Method.

Solution: in lecture.

Solution of Example 3.3-1 :

Write the LP model in standard form:
Max z = 5x1 + 4x3 + 0s1 + 0s2 + 0s3 +0s4

=⇒ z − 5x1 − 4x3 − 0s1 − 0s2 − 0s3 − 0s4 = 0

s/t 6x1 + 4x2 + s1 = 24

x1 + 2x2 + s2 =6

−x1 + x2 + s3 =1

x2 + s4 = 2

x1 , x2 ≥ 0

Institute: MGT, BUP Page# 37

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Simplex Tableau:

↓
Basic z x1 x2 s1 s2 s3 s4 Solution Ratio
z 1 -5 -4 0 0 0 0 0
24
← s1 0 6 4 1 0 0 0 24 6 =4 ← (min)
6
s2 0 1 2 0 1 0 0 6 1 =6
1
s3 0 -1 1 0 0 1 0 1 −1 = −1
2
s4 0 0 1 0 0 0 1 2 0 =∞
↓
Basic z x1 x2 s1 s2 s3 s4 Solution Ratio
z 1 0 − 23 5
6 0 0 0 20
2 1
x1 0 1 3 6 0 0 0 4 4 ÷ 23 = 6
4
← s2 0 0 3 − 16 1 0 0 2 2 ÷ 43 = 1.5← (min)
5 1
s3 0 0 3 6 0 1 0 5 5 ÷ 53 = 3
s4 0 0 1 0 0 0 1 2 2÷1 = 1
Basic z x1 x2 s1 s2 s3 s4 Solution Ratio
3 1
z 1 0 0 4 2 0 0 21 Optimum Soln
1
x1 0 1 0 4 − 12 0 0 3
x2 0 0 1 − 18 3
4 0 0 3
2
3
s3 0 0 0 8 − 54 1 0 5
2
1
s4 0 0 0 8 − 34 0 1 1
2

⇒ Based on the optimality condition, none of the z-row coefficients are

negative. Hence, the last tableau is optimal.

⇒ The optimal values of the variables in the Basic column are given in
the right-hand-side Solution column and can be interpreted as

Institute: MGT, BUP Page# 38

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Decision Var. Optimum Value Recommendation

x1 3 Produce 3 tons of exte-
rior paint daily
x2 1.5 produce 1.5 tons ffo in-
teror paint daily
z 21 Daily profit i s $21,000

⇒ The solution also gives the status of the resources.

Resources Slack Value Status

Raw material, M1 s1 = 0 Scarce (used the re-
source completely)
Raw material, M2 s2 = 0 Scarce
Market limit s3 = 2.5 Abundant
Demand limit s4 = 0.5 Abundant

Remark:
In minimization problems, the optimality condition calls for selecting the
entering variable as the nonbasic variable with the most positive objective
coefficient in the z-row, the exact opposite rule of the maximization case.
This follows because max z is equivalent to min (−z ). As for the
feasibility condition for selecting the leaving variable, the rule remains
unchanged.

Institute: MGT, BUP Page# 39

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
Example : From Hillier pp.94
Graphical Solution:

Institute: MGT, BUP Page# 40

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

The Simplex Method:

Institute: MGT, BUP Page# 41

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

with c2 = 2 instead of 5:

Institute: MGT, BUP Page# 42

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

Institute: MGT, BUP Page# 43

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
Example : From Gupta pp.166
LP Problem:

Standard LP Problem:

The Simplex Tables:

Institute: MGT, BUP Page# 44

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain

3.4 Artificial-Variable Technique

As demonstrated in Example 3.3-1, LPs in which all the con-

straints are (≤) with nonnegative right-hand sides offer a conve-
nient all-slack starting basic feasible solution.

Models involving (=) and/or (≥) constraints do not. The proce-

dure for starting “ill-behaved” LPs with(=) and (≥) constraints is
to use artificial variables that play the role of slacks at the first
iteration.

The artificial variables are then disposed of at a later iteration.

Two closely related methods are introduced here:

– the Big M-method and

– the Two-phase method.

Institute: MGT, BUP Page# 45

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
3.4.1 Big M-Method

Example form Hillier pp.117

Consider the following LP with two variables:

Max z = 3x1 +5x3

s/t x1 ≤ 4

2x2 ≤ 12

3x1 + 2x2 = 18

x1 , x2 ≥ 0

Solve this LP using the Big-M Method.

Solution: in lecture.

Remark:
⇒ Artificial variable objective function coefficient

 −M, in maximum problems
=
 M, in minimum problems

⇒ M isa sufficiently large positive value (mathematically, M → ∞)

3.4.2 Two-Phase Method

ˆ In the M-method, the use of the penalty, M, can result in computer
roundoff error.

ˆ The two-phase method eliminates the use of the constant M altogether.

ˆ As the name suggests, the method solves the LP in two phases:

Institute: MGT, BUP Page# 46

MGT4202: System Modeling & Decision Making © Dr. A B M Shahadat Hossain
(i) Phase-I attempts to find a starting basic feasible solution, and
(ii) if one is found, Phase-II is invoked to solve the original problem.

Example from Gupta, pp.192

Use the two-phase simplex method to solve the following LP:

Minimize z = 5x1 + 3x2

s/t
2x1 + x2 ≤ 1

x1 + 4x2 ≥ 6

x1 , x2 ≥ 0

Solve this LP using Two-Phase Method.

Solution: in lecture.

Institute: MGT, BUP Page# 47

Bibliography

[1] Hamdy A. Taha, Operations Research An Introduction, 10th Edition,

Pearson, 2017.

[2] Frederick S. Hillier and Gerald J. Lieberman, Introduction to Operations

Research, 10th Edition, McGrawHil Education, 2015.

[3] Prem Kumar Gupta, D. S. Sharma, Operations Research, S. Chand &

Company Pvt. Ltd., 7th Revised Edition, 2014.

48